racer

Every day you can read about how a company is using “Big Data” to get at the world’s “Big Problems.” I’ve always been dubious of these kinds of pitches, because the concept itself is so nebulous. Big data, at least to me, just means that we have a lot more documentation about what humans are doing. Yes, we can catalog and regress more data about their behavior, but that doesn’t necessarily mean we will actually reach any real insight.

But while there’s this big push to harness “Big Data,” some, like Bay-area-based Lumosity, are quietly exhibiting new and profound methods to take this mass of numbers to new analytic levels.

Lumosity makes brain training and memory exercises that people can perform in the comfort of their own home or on their mobile device. The games are fun brain teasers aimed at increasing users’ cognitive abilities and memory.

While users play Lumosity’s games, the company compiles each data point of their users’ performances, and plows this data into its own cognitive research projects. The company touts more than 50 million users and this large customer pool allows the company to create an enormous subset of participant data. This helps Lumosity analyze which programs work best, and even contribute real scientific findings to the neuroscience community.

This past weekend the company unveiled its most recent study, which tracked users’ learning rates through changing parameters. In essence, it showed that the researchers were able to alter certain games to maximize users’ ability to learn. Lumosity’s scientists looked at how participants played certain games and changed facets within the games’ structure to see if they increased participants’ ability to master the games. The scientists focused on two memory and response games.

Aaron Kaluszka, one of Lumosity’s data scientists, described one of the games as a “response inhibition task.” It placed a flock of birds on the users’ screens and the player would have to spot an opposite-facing bird. Users had to look at the mass of birds, spot the target bird, and ignore the other birds.

Kaluszka explained that the researchers changed the size of space between each bird for some users, as well as the sizes of the birds themselves, to see if there was a certain set of parameters that increased participants effectiveness at winning the game. The results found that when the parameter’s of the birds’ rotation were altered in a certain way, participants became more effective at mastering the exercise.

The study also looked at a memory game. Users were presented with a smattering of individual tiles for a few seconds and asked to replicate the array. If successful, the number of tiles would increase the next round.

This test, according to Kaluszka, produced the most interesting results. The researchers changed the rules behind the game: Originally if someone made a mistake the number of tiles would decrease. The study allowed some participants to continue responding even if they got a tile wrong, and this significantly boosted users’ results. Using this method, “people [could] over time and more rapidly increase the amount that they could remember,” Kaluszka told me.

In essence, alterations to these small parameters — be they rules or the size between objects — greatly affected the rates at which users learned these games.

This is another example of the way Lumosity is harnessing its immense data to refine its program. I wrote before about how the ways that Lumosity’s big data show possibilities for new kinds of scientific advancements. My editor, Adam Penenberg, wrote about a study that found that breast cancer patients who had undergone chemotherapy showed “significant” improvement in cognitive flexibility, verbal fluency, and processing speed after playing the company’s games four times a week for 12 weeks.

This latest study adds more grist to the power of games mill. Before, neuroscientific research was performed on a small subset of individuals, which caused larger percentage errors and less studies being performed. With a dataset like Lumosity’s, researchers can use its millions of data points to more easily finds scientific correlations.

It could create a sea change in scientific methodology, at least for the kind of research programs like Lumosity depend on.

In the future, if game players continue to pump data into the system, this could mean research that could alter pedagogy. Currently it’s only bird games, but who knows what could come next.

[Image via Thinkstock]